The OR Society’s Criminal Justice Sector Special Interest Group held a meeting this week and we had four really interesting and diverse presentations. Nick Malleson and Nick Addis talked about their work on the prediction of burglaries using Agent Based Modelling (ABM). They explained that it enabled them to model crime at the level of individuals with a type of crime where lots of good data are available. ABM Simulations focus on the actions and interactions of autonomous, interacting agents (burglars and their victims). Their model used real geographic data for a part of Leeds and were able to make comparisons between the outputs of the model and actual crime data. Nick Addis is building on this work as part of his PhD. For more information on crime simulation, visit Nick Malleson’s blog. Richard Thorne from NOMS spoke about Offender Segmentation and how his team had been building a model to help with the annual commissioning of non-core offender services. The heart of the segmentation model was the use of NOMS OGRS data. These data are actuarial predictions of the risk of re-offending. The model had been useful in opening up conversations with key stakeholders such as Prison Governors, NOMS Commissioners, Psychologists and others within the NOMS analyst community. The presentation prompted some challenging questions from Probation colleagues among the attendees, who pointed out that risk of re-offending is only one element of an individual offender’s risk profile. Ian Seed from Cogentus Consulting who spoke about a project to measure the effectiveness of a community-based police information tool being used in Los Angeles. The work was initiated because there was a concern that the information system was not being used as much as it could be by police officers on the beat. the project involved a series of stakeholder workshops to identify gaps/issues and to prioritise those. The Triz problem solving technique was then used to work through viable solutions in preparation for a Business Case to enhance the current system. Our final presentation was by Chris Polden of ORH Consulting who described a project to create a 20 year Facilities Plan for a police force. This looked at the current and projected demand for police services from the public in a City, divided into a number of districts. A travel time model was constructed to look at how travel time for members of the public to police stations could be minimised and how travel time for police officers could be reduced by optimally locating district police stations. One of the really interesting elements of the project was the constraints put on the model by the client; for example they placed constraints on changes in some working practices and some facility locations. This was a good demonstration of the “real world” where opinions and biases run into what a model might show as being an “improvement”. Slides from the session will be made available at the SIG’s web-page. Our next meeting will be on Monday 18th November, in London. Details to follow.